Show simple item record

Proc. IEEE ICASSP--10

dc.contributor.authorHuang, Songfang
dc.contributor.authorRenals, Steve
dc.date.accessioned2010-12-20T15:27:43Z
dc.date.available2010-12-20T15:27:43Z
dc.date.issued2010en
dc.identifier.urihttp://dx.doi.org/10.1109/ICASSP.2010.5495007en
dc.identifier.urihttp://hdl.handle.net/1842/4553
dc.description.abstractWe present an approximation to the Bayesian hierarchical Pitman-Yor process language model which maintains the power law distribution over word tokens, while not requiring a computationally expensive approximate inference process. This approximation, which we term power law discounting, has a similar computational complexity to interpolated and modified Kneser-Ney smoothing. We performed experiments on meeting transcription using the NIST RT06s evaluation data and the AMI corpus, with a vocabulary of 50,000 words and a language model training set of up to 211 million words. Our results indicate that power law discounting results in statistically significant reductions in perplexity and word error rate compared to both interpolated and modified Kneser-Ney smoothing, while producing similar results to the hierarchical Pitman-Yor process language model.en
dc.titlePower Law Discounting for N-Gram Language Modelsen
dc.typeConference Paperen
dc.identifier.doi10.1109/ICASSP.2010.5495007en
rps.titleProc. IEEE ICASSP--10en
dc.date.updated2010-12-20T15:27:43Z


Files in this item

This item appears in the following Collection(s)

Show simple item record